summaryrefslogtreecommitdiff
path: root/solenv/gbuild
AgeCommit message (Collapse)Author
2016-09-29gbuild: try to avoid invoking gdb with LD_LIBRARY_PATH setMichael Stahl
Commit b078e0d638514b298616838fae090b8294bac32f introduced the problem that gdb will load via LD_LIBRARY_PATH the libpython3.5m.so.1.0 from INSTDIR instead of the one from the system, which makes it unhappy (at least on Fedora 24). Try to hack around that by detecting that the user wants to run gdb, and passing in a command "set environment LD_LIBRARY_PATH ..." so that "run" inside gdb uses the custom LD_LIBRARY_PATH but gdb itself does not. Change-Id: I1d29d2af1a77ffab3e718a876c17c29de444c6d9 Reviewed-on: https://gerrit.libreoffice.org/29349 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Michael Stahl <mstahl@redhat.com>
2016-09-24show correct command to execute the testMarkus Mohrhard
Change-Id: Ic7515c1d43737fb6864228d37be1b356acecfdc3 Reviewed-on: https://gerrit.libreoffice.org/29253 Reviewed-by: Markus Mohrhard <markus.mohrhard@googlemail.com> Tested-by: Markus Mohrhard <markus.mohrhard@googlemail.com>
2016-09-22gbuild: remove gb_CFLAGS/gb_CXXFLAGS += -fno-omit-frame-pointerMichael Stahl
This did not take effect anyway because unxgcc.mk unconditionally overwrites it; see https://lists.freedesktop.org/archives/libreoffice/2016-September/075218.html Nobody knows why it was added so just remove it. Change-Id: Ib1a6d06ed8e380e63c588a72b6a65bfbeaacb924
2016-09-22gbuild: remove gb_CXXFLAGS += -mminimal-toc on PPC64Michael Stahl
This did not take effect anyway because unxgcc.mk unconditionally overwrites it; see https://lists.freedesktop.org/archives/libreoffice/2016-September/075218.html Change-Id: I33a2ae9aae44d974d6b5466a2d7dae61ae5cc92c
2016-09-22gbuild: gb_COMPILERDEFAULTOPTFLAGS doesn't existMichael Stahl
... since commit 8541943b8d08f62217e99c9d47a5f222e1fba86e Change-Id: I56b12ebd59b8387878f20d625381d3f743558a43
2016-09-22Use 'with open(file) as f:' to avoid 'ResourceWarning: unclosed file'Mike Kaganski
Removes 220 warnings from build log (at least under Windows). Requires Python 2.5+ Change-Id: I06b18b14a08f2c27528fcb009faddff081ffd54e Reviewed-on: https://gerrit.libreoffice.org/29172 Reviewed-by: Michael Stahl <mstahl@redhat.com> Tested-by: Mike Kaganski <mike.kaganski@collabora.com>
2016-09-22Fix sh quoting of ; and \ (the latter in the cygpath -w output)Stephan Bergmann
Change-Id: Iec52b93746b2fd9bbd6ad67667092717f39222d7
2016-09-22Make workdir/UnpackedTarball/langtag/data/.libs/lt-reg2xml find its libsStephan Bergmann
...after ecd5d49dc30d11845ccc1a14e26bf37d73959463 "liblangtag: prevent libtool from adding annoying stuff to RPATH". "Outsmarting" libtool in the way described in 1d028d4783da69c5c0e6e0b59e0f8ac55eb9d2b1 "Fix Linux RPATH of various external modules" only works if there's no utility applications in the outsmarted module, that are run while building the outsmarted module (cf. the redland case in 1d028d4783da69c5c0e6e0b59e0f8ac55eb9d2b1). To work around that, pass in an LD_LIBRARY_PATH (or equivalent) that finds the relevant libs as built in that module. (And gb_Helper_extend_ld_path was until now only available in com_GCC_defs.mk. There, it accepts any number of paths separated by ":". When used in a platform-independend way like it is used here, it will only support a single path.) Change-Id: Ic615e436b19b8b9aa7daa019c8f5287de90c3abc
2016-09-20Remove HAVE_GCC_VISIBILITY_BROKEN, never trueStephan Bergmann
The check for broken -fvisiblity-inlines-hidden doesn't mention a specific bug. Its roots date back to 072e89cf68dcf9b6713945e402dd7a9fe5cea48d "INTEGRATION: CWS newportstl" of the --with(out)-stlport area. Lets assume whatever failure is long since fixed. Make this a fatal configure error for now. The check should be removed completely after LO 5.3 branch-off. Change-Id: I7c64c2e8b9bd342beec9b84ff71f5712a6155b7d Reviewed-on: https://gerrit.libreoffice.org/29075 Reviewed-by: Stephan Bergmann <sbergman@redhat.com> Tested-by: Stephan Bergmann <sbergman@redhat.com>
2016-09-20[API CHANGE] Remove HAVE_GCC_VISIBILITY_FEATURE, always trueStephan Bergmann
* GCC documents -fvisibility at least as far back as the GCC 4.0 online documentation at <https://gcc.gnu.org/onlinedocs/gcc-4.0.4/gcc/Code-Gen-Options.html>. * For external code, odk/settings/settings.mk unconditionally set HAVE_GCC_VISIBILITY_FEATURE for all platforms other than Windows. Make this a fatal configure error for now. The check should be removed completely after LO 5.3 branch-off. Change-Id: I1de415b6ed1591e0a7b6640ece861b6f0ef74112 Reviewed-on: https://gerrit.libreoffice.org/29073 Reviewed-by: Stephan Bergmann <sbergman@redhat.com> Tested-by: Stephan Bergmann <sbergman@redhat.com>
2016-09-16Either -O2 is broken in Apple Clang, or some of our code is brokenTor Lillqvist
Already with Xcode 7.3.1 some unit tests crash when the code is built with full optimisation. Xcode 8 did not help. So use -O1. Yeah, this sucks. Don't have time to investigate in detail. Change-Id: I0cffad1a3ee3ee86b61eca3c856476133125049b
2016-09-12Disable GCC 6 -Wnull-dereference again for nowStephan Bergmann
...at least with both gcc-6.1.1-3.fc24.x86_64 and a recent GCC 7 trunk build it produces lots of incomprehensible (potentially false positive) occurrences in a --disable-debug build, like > [CXX] store/source/stordir.cxx > In file included from /home/sbergman/lo/core/store/source/stordir.hxx:30:0, > from /home/sbergman/lo/core/store/source/stordir.cxx:20: > /home/sbergman/lo/core/store/source/storbase.hxx: In member function ‘storeError store::OStoreDirectory_Impl::create(store::OStorePageManager*, rtl_String*, rtl_String*, storeAccessMode)’: > /home/sbergman/lo/core/store/source/storbase.hxx:237:23: error: potential null pointer dereference [-Werror=null-dereference] > m_nUsed = rhs.m_nUsed; > ~~~~^~~~~~~ > /home/sbergman/lo/core/store/source/storbase.hxx:236:23: error: potential null pointer dereference [-Werror=null-dereference] > m_nSize = rhs.m_nSize; > ~~~~^~~~~~~ > /home/sbergman/lo/core/store/source/storbase.hxx:235:23: error: potential null pointer dereference [-Werror=null-dereference] > m_nAddr = rhs.m_nAddr; > ~~~~^~~~~~~ > [...] > [CXX] store/source/storlckb.cxx > In file included from /home/sbergman/lo/core/store/source/storlckb.hxx:29:0, > from /home/sbergman/lo/core/store/source/storlckb.cxx:20: > /home/sbergman/lo/core/store/source/storbase.hxx: In member function ‘storeError store::OStoreLockBytes::readAt(sal_uInt32, void*, sal_uInt32, sal_uInt32&)’: > /home/sbergman/lo/core/store/source/storbase.hxx:595:26: error: potential null pointer dereference [-Werror=null-dereference] > return isA<U>(p) ? static_cast<U*>(p) : 0; > ~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~ <https://gcc.gnu.org/bugzilla/show_bug.cgi?id=16351> "NULL dereference warnings" looks like this is a known problem. Change-Id: I5eb0c9def7dd12e3a440d280e3fa2331317031b4
2016-09-05gbuild: remove hardlink deliverThorsten Behrens
Since there's hardly any 'deliver' going on anymore... Change-Id: Id349958d299c9742382d37deab2125bc09930c14 Reviewed-on: https://gerrit.libreoffice.org/28655 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: David Tardon <dtardon@redhat.com>
2016-08-30Fix placement of EXTRA_ENV_VARSStephan Bergmann
Change-Id: Id9ae15aaf24b65abfcb4006490836eae8a1d8db6
2016-08-30Hack for shaky VCL test under ASanStephan Bergmann
Change-Id: I0c89d1c23ddc1d0d6cd98d95ef31622a30d4862b
2016-08-18screenshots: add new global make targetArmin Le Grand
Up to now the screenshot creation was added/dependent of target slowcheck. Since quite some modules have added screenshot creations now, I added an own target 'screenshot' to allow to keep current slowcheck and screenshot creation separated Change-Id: I80a49a0db607edf8e0405672d570f624d29912e7
2016-08-18screenshots: unset VCL_HIDE_WINDOWSKatarina Behrens
Those screenshot cppunit tests need to run in non-headless mode with dialog windows exposed (i.e. VCL_HIDE_WINDOWS must not be set) To preserve the current behaviour (VCL_HIDE_WINDOWS is set always on Win and Mac, and in use_vcl_non_headless case on Linux), introduce 2 new variables: gb_CppunitTest_show_windows: unset VCL_HIDE_WINDOWS unconditionally gb_CppunitTest_hide_windows: set VCL_HIDE_WINDOWS=1 if HEADLESS is empty (i.e. let's not use emptiness of HEADLESS as an indicator, as it is empty in 2 mutually exclusive cases) Change-Id: Ib2f7a9cfb1202944d10856c44b6ac7c41156b333
2016-08-17enable password export tests under macosxCaolán McNamara
gerrit says they build with this cert folder disable variable in place Change-Id: Ibf711a01e6c055613c65dd94d85b523e7c8b5acd
2016-08-17"make debug=t" enables debuginfo againStephan Bergmann
...which had inadvertently been broken with 99db9f2295eb9a8b3288df9798a292b8d6e1b854 "Make --enable-symbols orthogonal to --enable-debug/-dbgutil" Change-Id: I3962aa8c67426f3aebc5ad746f7ac281c68d941a
2016-08-15Drop leftover EXTRA_CFLAGS and EXTRA_CDEFSTor Lillqvist
Change-Id: I006d19498f4b5c3bd4088fd1e290c30e7022b1c9
2016-08-05(initial) sparc64 portJames Clarke
Change-Id: I8ec9bb5415a9e6b9083ba89a7790872d67625de1
2016-08-02Remove some presumably unnecessary leftovers from ancient timesTor Lillqvist
I could not find that SHORTSTDC3 or SHORTSTDCPP3 would have any meaning. They only occur here (and in our history). Change-Id: I45141e60c144207fbf3fc0c5e59340b80951b212
2016-07-21gbuild: always set CCACHE_CPP2=1 when COM_IS_CLANGMichael Stahl
There are not only spurious warnings from loplugin, but also from plain clang, at least with the Fedora 24 version. So let's just try to always set CCACHE_CPP2=1. This should make the configure check for that obsolete. The check didn't do anything on Fedora anyway because it sets up ccache by putting wrappers like /usr/lib64/ccache/clang on PATH. Change-Id: Ieb17be4a1b1fd0f7aedb16b18cb49015917a26d3 Reviewed-on: https://gerrit.libreoffice.org/27345 Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com> Tested-by: Norbert Thiebaud <nthiebaud@gmail.com>
2016-07-21gbuild: set CCACHE_CPP2 automatically with --enable-compiler-pluginMichael Stahl
The plugins are confused about source file locations without CCACHE_CPP2, so try to avoid having to set it manually. Change-Id: I8c19cb1b4bea27963b2f416faa67a70c6c65da38 Reviewed-on: https://gerrit.libreoffice.org/27344 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
2016-07-11Fix --enable-mergelibs build.Jan Holesovsky
Change-Id: I839b9f8769508b8b4cc4e766db51188353767b0c
2016-07-11Make --enable-symbols orthogonal to --enable-debug/-dbgutilStephan Bergmann
Change-Id: I523bc1d848e40489370eefe00046e0a257ed2505 Reviewed-on: https://gerrit.libreoffice.org/27058 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
2016-07-11Get rid of odd ENABLE_SYMBOLS=FALSEStephan Bergmann
* None of the other ENABLE_* support anything but TRUE or empty. * Other checks for -n "$ENABLE_SYMBOLS" in configure.ac are confused by it. * solenv/gbuild/gbuild.help.txt talks about "If not empty", not mentioning "FALSE". * A missing (redundant) explicit --disable-symbols left gb_SYMBOL uninitialized (instead of setting it to $(false))---but without consequences, as all the places checking gb_SYMBOL explicitly check for $(true). Change-Id: Id004189de27a7936862dab3a5fd84d549c06af2b Reviewed-on: https://gerrit.libreoffice.org/27057 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
2016-07-11Break gb_DEBUGINFO_FLAGS out of gb_DEBUG_CFLAGSStephan Bergmann
...in preparation of making them orthogonal Change-Id: If75b334c954138b3aed4f8d1ac33061a2267ad52 Reviewed-on: https://gerrit.libreoffice.org/27056 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
2016-07-11Centralize setting gb_DEBUGINFO_FLAGS for gb_SYMBOL in LinkTarget.mkStephan Bergmann
Change-Id: Ie8ca63d48f66833a778342af8fbe19006fb6f143 Reviewed-on: https://gerrit.libreoffice.org/27055 Reviewed-by: Stephan Bergmann <sbergman@redhat.com> Tested-by: Stephan Bergmann <sbergman@redhat.com>
2016-07-10uitest: running uitests under gdb makefile partMarkus Mohrhard
Change-Id: If1e1d9217cfcc1f51fa611457ac7758fd2d9960b
2016-06-28gbuild: recognize newfangled screen TERM=screen.xterm-256colorMichael Stahl
Seems to have changed from Fedora 23, where it was just "screen", to 24. Change-Id: I9078c337093d12d62e9afe10b1f02aaf346d4b43
2016-06-24uitest: seems like the python_shell is not a necessary dependencyMarkus Mohrhard
Change-Id: Ia725af391dcfe6a1d8d2f79317ddfde241951d26 Reviewed-on: https://gerrit.libreoffice.org/26623 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Markus Mohrhard <markus.mohrhard@googlemail.com>
2016-06-24uitest: the correct python executable for windows is in instdirMarkus Mohrhard
Change-Id: Ica1efa92d6b3b26b3d138763c8f548c59d53e346 Reviewed-on: https://gerrit.libreoffice.org/26622 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Markus Mohrhard <markus.mohrhard@googlemail.com>
2016-06-22uitest: start the switch to python's unittestMarkus Mohrhard
Change-Id: Ic6e0d99a567b7080928accb61ed036b1c25ec7ff
2016-06-20switch to EHs on windowsMarkus Mohrhard
This seems to be a good idea based on several discussions in the project. In the end catching SEH exceptions is just going to cause strange platform dependent bahavior. This patch is based on on http://thread.gmane.org/gmane.comp.documentfoundation.libreoffice.scm/39102/focus=55516 and includes some additional cleanup of the sal signal code. Change-Id: Iedc998e37e6495afec445eccb60fa1c2b1a7defd Reviewed-on: https://gerrit.libreoffice.org/26497 Reviewed-by: Michael Meeks <michael.meeks@collabora.com> Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
2016-06-18uitest: build system part for new uitestsMarkus Mohrhard
We now can call the uitests with make uitest.uickeck Change-Id: I20c73efd93c7987f3b841cd0e3e7842ee7a5dab9
2016-06-15tdf#98955 hardware_concurrency not ideal for thread poolsAshod Nakashian
A new static member getPreferredConcurrency added to comphelper::ThreadPool to return a configurable max number of threads. By default the new function returns the hardware_concurrency value provided by std::thread. When MAX_CONCURRENCY envar is defined, the return value is limited to whatever is set there. Three call-sites that used std::thread::hardware_concurrency have been replaced with getPreferredConcurrency. Unittests added to cover the functionality of the new member. Unittests are capped to 4 threads. Change-Id: I3332e393a88a5ed436316fa712ed920a4b37f4af Reviewed-on: https://gerrit.libreoffice.org/26254 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
2016-06-10CygWin is also known terminalVasily Melenchuk
This will enable colors in build output in CygWin Change-Id: I88ba01c35320783a418a87a12cbb17c56000c396 Reviewed-on: https://gerrit.libreoffice.org/26153 Reviewed-by: Noel Grandin <noelgrandin@gmail.com> Tested-by: Noel Grandin <noelgrandin@gmail.com>
2016-06-07pass image lists using a response fileDavid Tardon
Change-Id: Ic00715edd1173ebb01ce5853e2835d71235fa4df
2016-06-03try to avoid 'by-design' solar-mutex deadlock during ciNorbert Thiebaud
Change-Id: I9e1d79613adf6184f76c2e07aca0b78a9329838f Reviewed-on: https://gerrit.libreoffice.org/25839 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
2016-05-27shortcut the signal handler during batch testNorbert Thiebaud
Change-Id: I7848555743e8d54e755dc82cbf0148eeed54110f Reviewed-on: https://gerrit.libreoffice.org/25178 Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com> Tested-by: Norbert Thiebaud <nthiebaud@gmail.com>
2016-05-25crash report lib should be part of mergelibMarkus Mohrhard
Change-Id: I23c338f9fcf7eeea4a440013ea6c1b9206e040f3 Reviewed-on: https://gerrit.libreoffice.org/25430 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Markus Mohrhard <markus.mohrhard@googlemail.com>
2016-05-20remove more linking noise from VisualStudio buildNoel Grandin
Change-Id: Id6254b9b261e61e567ce3a557fa8e6cb30cfc3ae Reviewed-on: https://gerrit.libreoffice.org/25155 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
2016-05-11Enable Clang -Wimplicit-fallthroughStephan Bergmann
...now that all intended ones are marked as SAL_FALLTHROUGH. Change-Id: Ibc76299cdac0d33285a47b106e7a548a5efffcbd
2016-05-05Fix linking of 64 bit artifacts in 32 bit Windows build with VS2015Tomofumi Yagi
In some 64bit Windows environments we get the link error while building libraries for 64-bit Explorer extensions. This patch will fix this error. This error occours only when we build 32bit LibreOffice on 64bit Windows with Visual Studio 2015. In some environments UCRT SDK directory does not match Windows SDK directory. In that case UCRT library's directory path is incorrect at the linking phase. Change-Id: I245b1a70f5e1137a8840aa2e7a70f521eab18cb3 Reviewed-on: https://gerrit.libreoffice.org/23982 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
2016-05-03Handle wchar_t as native C++11 type on windowsDavid Ostrovsky
The option /Zc:wchar_t- prevented to use wchar_t as a built-in type according to the C++ standard. In Visual C++ 6.0 and earlier, wchar_t was not implemented as a built-in type, but was declared in wchar.h as a typedef for unsigned short. Now, years later after the end of life this outdated toolchain, there is no reason not to use native type. The only issue could be the ABI compatibility. But on a quick look at least, it looks like none of the mangled C++ symbols in the stable URE interface actually depend on wchar_t. We forgot to get rid of /Zc:wchar_t- in 5.1. Do that for LibreOffice 5.2, though. Change-Id: I8d6b380660859efa44c83c830734978d31d756a0 Reviewed-on: https://gerrit.libreoffice.org/22589 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
2016-04-28Revert "add additional modules to merged lib"Oliver Specht
Breaks tinderbox build in cppumaker This reverts commit 3393e04edd0f4f6baf5faeda2ed761332f45d9e5.
2016-04-29add additional modules to merged libOliver Specht
acc, expwrap and store are loaded at startup anyway Change-Id: I28020962238be3fe73a1b1dbba59a45c7d3c1e1d Reviewed-on: https://gerrit.libreoffice.org/23459 Tested-by: Jenkins <ci@libreoffice.org> Reviewed-by: Oliver Specht <oliver.specht@cib.de>
2016-04-26Enable Clang plugin warnings in Bison source codeStephan Bergmann
-Werror is generally suppressed in Bison-generated C/C++ code (as in all other generated code) to silence warnings from the Bison skeleton code. And the Clang plugins suppress warnings in generated WORKDIR code based on the presumed source location (i.e., taking #line directives into account). So introduce a new PLUGIN_WARNINGS_AS_ERRORS mode where warnings from Clang plugins are reported as errors even if -Werror is suppressed. That way, any warnings in the Bison skeleton code still do not lead to compilation errors, while (at least plugin- emitted) warnings in the genuine source code do. Unfortunately this cannot also be enabled for Flex source code, as at least Flex 2.5.39 generates poor code that does not properly prefix all skeleton code with appropriate #line directives, so that some skeleton code would be mistaken for genunie source code, and compilation would fail due to errors. Also, %glr-parser Bison input appears to generate no #line directives at all (at least with Bison 3.0.4), so all of connectivity/source/parse/sqlbison.y is considered generated code and plugin warnings are still suppressed throughout. Change-Id: Id746e81cbfa5f77628b0a34c7b82780948e7db08
2016-04-18Enable further GCC 6 warningsStephan Bergmann
...not already automatically enabled through the existing -W switches. The set is assembled from information in <https://gnu.wildebeest.org/blog/mjw/2016/02/15/looking-forward-to-gcc6-many-new-warnings/>. All issues found have already been fixed with earlier commits. Change-Id: I1abb6ba3a12c0ffb2756d06476e395b24e596f8e Reviewed-on: https://gerrit.libreoffice.org/24219 Reviewed-by: Stephan Bergmann <sbergman@redhat.com> Tested-by: Stephan Bergmann <sbergman@redhat.com>
generated by cgit (git 2.34.1) at 2025-02-10 14:55:14 +0000